Volume 10, numéro 1, 1997

FR :

Au cours des dernières décennies, la hausse de la productivité agricole s'est accompagnée d'une forte augmentation des fertilisations azotées qui a entraîné l'augmentation des concentrations en nitrates dans les eaux souterraines. Récemment, les études sur la gestion des pollutions diffuses agricoles ont intégré l'échelle du bassin versant hydrologique. Dans cet article, une approche basée sur un découpage en secteurs pédologiquement et agronomiquement homogènes a été élaborée pour permettre l'utilisation d'un modèle d'évaluation (AgriFlux), et obtenir les flux d'eau et de nitrates sortant de la zone racinaire sur l'ensemble d'un bassin. La modélisation du bassin de La Jannerie a porté sur une période de quatre ans impliquant 19 zones de simulation. L'influence du cycle végétatif des cultures et leur nature sur l'évolution des flux de nitrates a été mise en évidence par des simulations préliminaires, de même que l'influence du type de sol. L'évolution des concentrations moyennes saisonnières en nitrates sortant de la zone racinaire montre que les fertilisations minérales ne sont pas les seules sources importantes de nitrates dans les sols. Les pratiques culturales, comme le retournement des prairies, l'enfouissement des résidus de récoltes ou l'assolement, ont une forte influence sur la dynamique spatiale et temporelle des flux de nitrates percolant vers la nappe.

EN :

In agricultural regions, groundwater contamination by nitrogen compounds originating from fertilizers is one of the most significant environmental problems. Along with in situ monitoring, simulation models have been developed for non point pollution (nitrates, pesticides) in order to evaluate both the level and the extent of the contamination. Simulation models, originally intended for research purposes in relation to the dynamics of agricultural systems, have been adapted and applied to environmental management in order to quantify water volumes and contaminant masses likely to reach groundwater systems. Recently, mechanistic models such as Agriflux (Banton et al., 1993) have been developed for use in the field. Agriflux is based on a mechanistic approach to the processes and incorporates a stochastic analysis that takes into account the spatial variability of the parameters. lt. calculates nitrate concentrations as well as water fluxes in the unsaturated zone. In the present study, environmental management principles integrating heterogeneity in soils and agricultural practices were applied to an agricultural watershed in Poitou (France). Preliminary simulations were carried out in order to estimate the influence of various parameters on the nitrate and water fluxes. First, a three-year wheat mono-crop was simulated using the same fertilization rate for each year. The calculated nitrate concentrations follow a trend opposite to that of the seasonal growing crop. To estimate the influence of the soil characteristics on the nitrate concentrations, the four types of soil in the watershed were simulated using the same three-year crop rotation. The results show that the soil type directly influences the amount of nitrate leaching. Under different soils types, the evolution of the concentrations over time follows the same pattern, but the concentration levels are significantly different. To quantify the impact of crops on the nitrate concentrations, the main crop rotations were simulated for the same type of soil. This set of simulations underlines the environmental differences between winter and spring crops. lt. also shows the differences induced by the presence of residues. The La Jannerie watershed was divided into homogeneous zones for soil and crop characteristics. During a four-year period, seasonal and annual nitrate concentrations were calculated for each homogeneous zone from the daily water and nitrate fluxes simulated with Agriflux. The results demonstrate the influence of the agricultural practices on the calculated concentrations. Overall, nitrate levels remain quasi-constant during the periods when the crops are active but vary considerably during the winter when the crops are absent or inactive. This winter period corresponds to a peak in nitrate leaching because of the excess rainfall and the absence of nitrogen uptake by the plants. The incorporation of crop residues in the soil in the autumn generates a high production of nitrates during winter due to the mineralization of the organic nitrogen.

Two different environmental approaches can be used jointly to evaluate agricultural practices. The first consists of a comparison between the nitrate flux that can reach the saturated area and the fertilizer rate. This approach provides an estimate of the amount of nitrogen lost to the aquifer. Simulations with Agriflux show that the nitrate fluxes are highest during the autumn when plant uptake is non-existent, except in fields with winter crops. The second approach compares the calculated nitrate concentrations that may occur in the aquifer with recognized water quality criteria. lt. is interesting and important to note that, during the simulated period, the calculated concentrations in the leach were often much lower than the water quality criterion (50 mg NO3/L). This result indicates that the fertilization practices applied in the watershed during this period tended to approach the real crop requirements (minimal requirements) and were more environmentally adequate (environmental optimum) than those used previously.

FR :

Le Canada possède l'une des plus importantes réserves d'eau douce au monde. La province de Québec n'y fait pas exception puisqu'on y trouve près de 750 000 lacs et une multitude de rivières. Une grande partie de cette ressource est très fragile et exposée à divers problèmes environnementaux. Ces problèmes prennent leur source à l'échelle locale, régionale, continentale ou même planétaire dans certains cas et sont de différentes natures : eutrophisation, contamination par diverses substances polluantes, acidification, changements climatiques, etc. Au Québec, les moyens d'action actuels visent principalement à réduire les pollutions à la source par voie de réglementation, de programmes d'incitation ou de programmes de subvention (actions d'assainissement des eaux), en ne recourant aux interventions directes à court terme que dans des situations d'urgence ou d'extrême nécessité. Le présent texte rend compte des problèmes liés à l'eutrophisation, à l'acidification et à la contamination, ainsi qu'à leur échelle d'occurrence et des solutions disponibles pour amoindrir leur impact. Cette discussion est menée en favorisant une approche globale concertée qui vise à établir les bases d'une réelle politique de gestion de l'eau.

EN :

Canada has one of the greatest freshwater reserves in the world. The province of Québec is not an exception to this generalization in that there are over 750 000 lakes and a large number of rivers of all sizes within the provincial boundaries. A major portion of this resource is highly vulnerable and is exposed to many contamination and environmental problems such as eutrophication, contamination from various pollutants, acidification, global climate changes, etc. In the province of Québec, actions are presently oriented toward reducing pollution hazards at the source by way of legislation, incentives and grant programs (wastewater cleanup programs), while resorting to short term direct approaches only in cases of emergencies and extreme necessity. The present article summarizes information about the problems encountered in Québec, their incidence, and the range of solutions available to reduce the impacts. This reflection is conducted in a way that favors a concerted global approach, which could act as the basis for a true water management policy.

FR :

Cette étude a consisté à évaluer et à comparer l'impact des traitements de coagulation-floculation, ozonation et filtration sur charbon actif en grains sur la matière organique dissoute de différentes eaux de rivières et de retenues françaises en s'appuyant sur le suivi de deux paramètres principaux, la fraction biodégradable du carbone organique dissous (CODB) et la distribution des substances humiques et non humiques (établie sur la base d'un protocole de filtration en série sur résines XAD-8 et XAD-4 ).

Dans le cas des eaux étudiées, la coagulation-floculation s'est accompagnée d'une diminution importante du COD, soit 38 à 70 %, impact qui se répercute dans des proportions équivalentes au niveau de sa fraction biodégradable soit 38 à 88%. Dans la majorité des cas, la clarification conduit à l'élimination préférentielle des substances humiques (définies comme hydrophobes), qui correspondent aux fractions de plus hautes masses moléculaires. Comparativement, et pour des taux supérieurs à 1 mg O3/mg C, l'ozonation entra"ne une réduction de la fraction des substances humiques qui se traduit par une augmentation de la fraction des substances non humiques et en particulier des substances hydrophiles non adsorbées sur résines XAD-8 et XAD-4. Cette modification, d'autant plus marquée que le taux d'ozone est important, s'accompagne d'une augmentation proportionnelle de la fraction biodégradable du COD. Le suivi en usine a montré que les taux d'ozone appliqués lors d'une interozonation modifient peu la nature du COD. Par contre, la filtration sur charbon actif en aval change de manière importante la distribution de la matière organique dissoute avec une augmentation relative de la fraction hydrophobe et des composés de faibles masses moléculaires apparentes (< 1 000 daltons).

EN :

The goal of our study was to evaluate the impact of water treatment processes (i.e. coagulation- flocculation, ozonation and GAC filtration) on the natural organic matter (NOM) of various river and reservoir waters based on DOC and BDOC analyses and the determination of the humic/non humic NOM distribution (fractionation of the DOC at acidic pH using two successive XAD-8 and XAD-4 resin columns). Analyses carried out on ten French raw surface waters have shown that the BDOC fraction accounted for 11 to 38 % of the DOC. The humic/non humic distribution of the NOM varied slightly with the origin of the studied water. About 50 to 60 % of the DOC was found to be incorporated into the humic fraction (NOM adsorbed on the XAD-8 resin), the hydrophilic acids (adsorbed on the XAD-4 resin) accounted for 10 to 25 % of the DOC while the non adsorbed hydrophilic solutes (hydrophilic neutrals that constitute the XAD-8/XAD-4 effluent) represented 15 to 30 % of the DOC.

Clarified water samples were collected from water treatment plants after coagulation/flocculation/sand filtration when no preoxidation was used. For water utilities which included a preoxidation step in their treatment process, raw water samples were coagulated and flocculated at a laboratory scale with Jar Test equipment using the same conditions (nature of the coagulant, pH, dose) as those used in the corresponding treatment plant. Globally, 38 to 70 % of the DOC and 38 to 88 % of the BDOC were removed during coagulation-flocculation, depending on the water site. In general, the humic/non humic NOM distribution of clarified waters showed a slight increase in the proportion of the non humic organic fraction as compared to raw waters, which indicates that humic substances (higher molecular weight organics) are preferentially removed during coagulation-flocculation.

Ozonation experiments were carried out on a raw water and a clarified water sampled from the same water site using a semi batch reactor (ozone was generated from high purity oxygen). For both waters, very little variation of the humic/non humic NOM distribution was observed for applied ozone doses around 0.5 mg O3/mg C or below. For higher ozone doses, the NOM distribution was dramatically changed despite only a small reduction of the DOC. The large reduction of the humic fraction was followed by a proportional increase of the non adsorbed hydrophilic solute fraction (small reduction of the DOC). As the applied ozone dose was increased from 1 to 3 mg O3/mg DOC, the non-adsorbed hydrophilic solute fraction also increased. The shift from high molecular weight organics such as humic materials to more hydrophilic organics (high polarity and low molecular weight organic solutes) during ozonation has often been mentioned in the literature. The increase of the non humic substances was followed by an increase of the BDOC. Results have also shown that higher ozone doses yield higher BDOC. Similar observations could be made with the raw and the clarified water.

As a conclusion of this work, samples were collected at the different steps of a water treatment plant (raw water, clarified water, intermediate ozonated water, GAC filtered water). The impact of coagulation/flocculation, and intermediate ozonation on BDOC and the humic/non humic NOM distribution confirmed the previous observations. The GAC filtration had a large impact on the DOC distribution while the BDOC was only slightly reduced. The NOM of the treated water was found to be more hydrophobic in nature with compounds that showed apparent molecular weights below 1 000 daltons (more than 80 % of the DOC).

FR :

Ces dernières années a été développé, au Cemagref du groupement de Lyon, un outil de synthèse (dit QdF) des régimes de crue des bassins versants, selon une approche multidurées et multifréquences des crues observées. QdF est donc un modèle continu de prédétermination des crues fréquentes à rares d'un bassin versant (observé ou non), qui permet de répondre à une conception de gestion intégrée des cours d'eau et de leurs bassins versants. Il nous a semblé utile, tout en se référençant aux principales publications et travaux de thèses, d'insister sur les notions de base et les concepts à l'origine des modèles QdF. Les variables hydrologiques étudiées, débit moyen maximal (VCXd) et débit seuil maximal (QCXd) sont relatives à une durée continue d pouvant varier d'une seconde à trente jours. Pour tout échantillon de durée d, constitué selon une technique d'échantillonnages de valeurs maximales indépendantes au dessus d'un seuil donné, l'adéquation de la loi exponentielle est en général vérifiée pour les fréquences observables (0.5²T(an)²20). Pour l'extrapolation aux fréquences rares (20<T(an)²1000) une forme théorique dite esthétique du GRADEX est privilégiée. Des lois théoriques, relatives aux observations et extrapolations, peuvent être ensuite déduites les courbes débit (Q)-durée (d)-fréquence (F) d'un bassin versant. Le choix de deux descripteurs du régime local qui renseignent sur la fonction de transfert (D) et sur la fonction de production (QIXA10) permettent de rendre adimensionnel le faisceau de courbes QdF précédent. A partir de ce dernier faisceau, il est possible de transférer sur un site quelconque, dont les descripteurs D et QIXA10 sont connus, des quantiles de crue pour toute durée et toute fréquence. L'opération est grandement facilitée par un formalisme mathématique (modèle QdF) de ce faisceau de courbes adimensionnelles. Les trois modèles QdF opérationnels actuels, chacun d'eux étant représentatif d'une famille hydrologique de bassins versants, ont permis lors de nombreuses applications de rendre bien compte de la grande variabilité spatio-temporelle des débits.

EN :

Using a multi-duration and multi-frequency description of observed flows, we have recently developed the Cemagref QdF model to represent watershed flood regimes. The model is also used for the predetermination of flood characteristics, on both gauged and ungauged watersheds, in the range of frequent to rare floods. By allowing upstream and downstream recalculation of flood characteristics, the model answers questions relevant to integrated river management. In the present article we have emphasized the basic notions and concepts underlying the QdF models; more detailed descriptions can be found in the referenced publications and theses. The studied hydrologic variables, maximal average flow (VCXd) and maximal threshold discharge (QCXd), are linked to continuous duration, d, which can vary from one second to thirty days. For every sample of duration d, composed with a sampling method of independent maximal values above a given threshold, the exponential law is generally adequate for mean return periods, for means in the range from 0.5 to 20 years (O.5²T(year)²20). A theoretical form, called "esthetic" GRADEX, is used to extrapolate observed floods to rare frequencies floods (20<T(year)²1000). From theoretical laws, relative to observations and extrapolations, discharge (Q) - duration (d) - Frequency (F) curves of a basin are then deduced. Using two basin descriptors, one for the transfer function (D parameter) and one for production (QIXA10 parameter), the preceding curves are then converted into an adimensional or normative form. Such dimensionless curves are then used to predetermine floods quantiles for all durations and frequencies, both on gauged and ungauged basins, using known site parameters D and QIXA10. Quantiles are easily computed from a mathematical formulation (QdF model) of normative QdF curves. Validation of QdF curves on about 250 basins revealed three main groups of hydrological flood regimes. Using the same mathematical form, the great diversity in flood characteristics in each group is well represented by one set of model parameters. The great variability of floods in space and time could then be predetermined in all basins, using three sets of fixed model parameters.

FR :

Une procédure d'évaluation régionale des risques de contamination des eaux souterraines par des pesticides a été développée et appliquée à une partie de la plaine du Rhône valaisanne. La combinaison d'une application stochastique (Monte–Carlo) de modèles déterministes simulant localement le devenir de pesticides et des techniques d'interpolation géostatistique permet d'évaluer également les incertitudes entachant les prédictions effectuées. Les divers types de modèles utilisés (solution analytique et résolution numérique de l'équation de convection-dispersion, modèle capacitif) conduisent en général à des résultats très similaires. Les cartes obtenues montrent que le risque de contamination est très élevé. Les incertitudes sont d'un ordre de grandeur similaire, i.e. ± 0.2-0.3 pour des indices de risque compris dans l'intervalle [0,1]. Ces incertitudes proviennent à raison d'environ 40-50 % des propriétés des pesticides et d'environ 30-40 % de la profondeur de la nappe phréatique, le 20 % restant étant dû aux incertitudes entachant les caractéristiques des sols, essentiellement leurs teneurs en carbone organique.

EN :

A method to assess groundwater vulnerability to pesticide contamination on a regional scale has been developed and applied to the Rhone river valley in Valais. Uncertainties regarding vulnerability predictions were accounted for by combining local stochastic simulations, i.e. one-dimensional deterministic pesticide fate models used in a Monte–Carlo process, and interpolation by geostatistical tools. Due to the characteristics of the dataset, several preliminary operations were required including:

- a statistical analysis of soil sample data to derive a probabilistic relationship between soil classes and hydrodynamic properties,

- an estimation of the relevant water table depths at soil profile locations by cokriging the minimal and maximal water table heights measured in neighboring wells,

- and an analysis of the climate data provided by two stations located close to the study area.

In the Monte–Carlo process, soils were described by the probability density functions of particle size distribution and organic carbon content as derived for each soil class from soil sample data and by an empirical uniform distribution of dispersivity. The parameters of moisture retention curves and hydraulic conductivity versus water content relationships were derived using various pedotransfer functions. Water table depth was described by a uniform distribution within the range defined by its minimum and maximum values while empirical uniform distributions outlined the properties of three generic pesticides. The important spatial variations in climate along the valley were accounted for by linear interpolation between the data from the two stations. The three selected simulation models were an analytical solution of the convection–dispersion equation (attenuation factor), a tipping bucket model (Leach–A) and a numerical solution of the convection–dispersion equation (Leach–M). Model sensitivity analysis using Latin hypercube sampling along with multiple regression showed that pesticide properties (degradation rate, partition coefficient), organic carbon content and water table depth are the most important variables regarding cumulative (10 year) pesticide fluxes to the groundwater. This analysis also stressed the weak effect of soil hydrodynamic characteristics. Besides, the Latin hypercube sampling technique proved to be very effective in reducing the number of simulations required by the Monte–Carlo process to something manageable. The five simulated cases were:

1. all uncertainties included,

2. fixed pesticide properties,

3. fixed pesticide properties and water table depth,

4. same as 3 but with empirical distributions for soil characteristics

5. same as 3 but with constant average climate conditions.

In all cases, the fractiles of the locally simulated distributions show the same spatial pattern, i.e. all their variograms and cross variograms are proportional to the same model. Due to this intrinsic coregionalization property, interpolation may be achieved by kriging independently the various fractiles rather than by cokriging simultaneously all of them. However, integration of spatial uncertainty by sequential gaussian simulation has not been achieved due to excessive computer lasting. The resulting maps show that groundwater vulnerability is very high. Uncertainties are almost of the same order of magnitude, i.e. ± 0.2 for vulnerability indices ranging within the [0, 1] interval. Uncertainties on pesticide properties and water table depth account each for some 40 % of the resulting uncertainty while the variability in organic carbon content accounts for the remaining 20 %. All model outcomes were quite similar, except in the case of the less persistent pesticide with the attenuation factor. The dominant variables are the degradation rate and the partition coefficient of the pesticide, the organic carbon content of the soil and water table depth. Climate is of course the governing process, but 10 to 20 % variations do not significantly affect the predicted cumulative pesticide fluxes. Availability of soil sample data regarding particle size distribution as well as pedotransfer functions has little effect on the resulting vulnerability assessments.

FR :

Une règle de gestion interannuelle empirique est définie pour gérer un barrage en éte d'un périmètre d'irrigation en zone semi-aride. La structure de la règle est construite à partir de l'observation de la gestion pratiquée actuellement sur le terrain. Les paramètres de cette règle sont ajustés de façon à minimiser le cumul moyen des écarts entre demandes et ressources des séries générées après une étude fréquentielle des aléas hydrauliques.

Dans le cas du périmètre lié au barrage de Ghézala, cette méthode permet de répartir au mieux les risques de défaillance sur l'année et d'adapter la stratégie d'irrigation aux phases végétatives critiques des cultures.

Les avantages et les inconvénients de cette approche empirique par rapport aux techniques de programmation dynamique sont finalement discutés et mis en relief pour les systèmes de gestion des ressources en zones semi-arides.

EN :

An empirical inter-annual management rule is defined to operate a reservoir for irrigation use in a semi-arid zone. Management rules for hydro-agricultural systems in semi-arid zones consist of establishing a trade-off between the total quantity of water to be distributed for summer irrigation and the indispensable volumes to be held in store for the raising of future seedlings during the following autumn. In less arid countries, water is sufficiently cheap so that it is economically profitable to deliver great quantities to sustain maximum agricultural output. Indeed, water is sufficiently abundant so that it is generally useless to conserve reserves for the next year. Such an inter-annual compromise is not the norm for water resources managers in humid climates. For instance, the two consecutive dry summers in Europe in 1989 and 1990 would not have had the same impact on the agricultural economy in the semi-arid zones. There rules of management would have been adapted to obtain of an inter-annual balance.

Numerous mathematical methods have been used in the domain of water resource systems management, sustained by the constant increase of computer performances. Reservoir management issues have consequently been widely studied and solved both from the viewpoint of the art of modeling and the methods of optimization. Nevertheless it is necessary to ensure that the mathematically correct optimal solution is also implementable when one leaves the computer screen to get back to the field: the success of this passage depends largely on the ability of the objective function to quantify the economic stakes encoded in the choice of a management rule. Except for hydropower generation, the various uses of a dam are most unlikely to be integrated into an aggregative utility function under the form of separable additive criteria.

A way to bypass the inherent difficulties in the design of a realistic objective quantification is to restrict the search of the operation rule to the class of strategies compatible with the commonly met attitude of semi-arid water resources system managers. The structure of the rule is designed from the observation of the effective management presently in operation. Its logic fits to the following reasoning:

- when the reservoir level is low, as soon as it starts to vary, the irrigation perimeter manager will behave very prudently. i.e. by reducing the proportion of the allocated irrigation water to the amount in store;

- for average working conditions, the system manager will deliver a constant fraction of the available reservoir storage for irrigation;

- when the quantity stored in the dam is high, the manager tends to satisfy completely the irrigation demand.

The parameters of this piecewise linear rule are adjusted so as to minimize the probability of discrepancy between demands and resources on series obtained after a brief hydrological study.

The Ghezala perimeter in Tunisia, which is situated in the watershed of lake Ichkeul, has been taken as a case study. With a thousand hectares of irrigated crops to be satisfied from an upstream 12 hm3 reservoir, and 625 mm rain per year on the watershed with a strong seasonal dependence, the Ghezala perimeter is quite representative of Tunisian irrigation practices. The method developed in this paper allows to balance the risk of failure all over the year and to adapt the strategy of irrigation according to critical vegetative phases of the crops. A statistical analysis of possible deficits according to the improved rule reveals that the method improves the performance of irrigation and decreases its sensitivity to the critical vegetative phases that are of critical importance in agriculture in semi-arid zones.

The management model is unsophisticated but its objective is to provide elements of appreciation to the manager of the irrigated perimeter. Each component of the model can be improved and more finely described. Advantages and drawbacks of this empirical approach with regards to more conventional dynamic programming techniques are finally discussed and underlined for water resource systems with specific attention to their management in semi-arid zones.